The Pacific giant salamander (Dicamptodon tenebrosus) is a vitally important member of the cool mountain stream ecosystems of the Washington and Oregon Cascades and coastal regions. Reaching up to thirteen inches in length, these semi-aquatic creatures depend on clean aquatic habitats for survival (Oregon Wild). With the current threat of logging and other anthropogenic activities in the region, the salamander’s watershed ecosystem has been at risk. Although not particularly threatened, the Pacific giant salamander is sensitive to some of these changes, specifically small changes in water quality. Given the current anthropogenic activities and their risks of stream siltation and higher stream temperatures, it is essential to analyze data on the salamander to assess the overall health of the watershed ecosystem.
Figure 1. Pacific giant salamander (Credit: Sierra Nystrom)
By measuring physical trends related to the species, along with land use and disturbance history, we can begin to glean the effects of the aforementioned human activity. Hopefully by doing this, recommendations can be made on how to proceed with logging and conservation in these regions.
Data for Pacific giant salamnder numbers and size from 1993-2017 was provided by the Aquatic Vertebrate Population Study conducted within Mack Creek in the Cascade Mountains of Oregon. Mack creek is located within the A.J. Andrews Experimental Forest which is part of the Long-Term Ecological Research (LTER) Network. Founded in 1948 by the US Forest Service, the forest is the center for forest and stream ecosystem research in the Pacific Northwest. Beginning in the 1950s, several small watersheds were manipulated (for example, logged or not logged) to lay a foundation for research on how the forest and streams interact (Andrews Experimental Forest). For more information on collection methods and spatial information, see Metadata. This analysis used chi-squared testing to compare differences between salamander counts. Homogeneity of variances was done by a Levene’s test. Mean salamander weights in 2017 were compared using a two-sided t-test for two groups (old growth vs. clear cut reaches of Mack Creek), and one-way ANOVA with post-hoc Tukey’s when comparing >2 groups (channel classification); α = 0.05 was used throughout. Effect sizes were calculated using Cohen’s d. All analyses and figures were prepared using R software version 3.6.1.
Figure 2. The Mack Creek study site is located in the HJ Andrews Experimental Forest, Oregon. Salamander data for this analysis was collected from the area within the red square. (Credit: Dana Warren)
Figure 3. Mack Creek, HJ Andrews Experimental Forest, OR. The landscape is steep, with hills and deep valleys. Cold and fast running streams flow through the many valleys of the forest, most of which consists of dense old-growth tree species. (Credit: Oregon State University)
Since these salamanders can be sensitive to changes in water quality and temperature it is important to visualize the differences in salamander abundance between the section of Mack Creek that runs through old-growth forest and the section that runs through clearcut forest. Forest logging can accelerate erosion and opens up the forest floor to increased solar radiation, both of which may affect the salamander’s habitat.
Figure 4. Number of Pacific giant salamanders recorded in Mack Creek from 1993-2017. Salamander counts have been seperated by whether they were recorded in the section of creek that runs through old-growth forest or forest that was clearcut. Data: Andrews Forest LTER
We can see that trends in salamander abundance in both sections of the creek closely mirror one another from 1993 to 2017. In general, the abundance of Pacific giant salamander is greater in the old-growth section of Mack Creek. However, in 2003 salamander counts are nearly equal and in 2014 abundance within the clearcut section of the creek becomes greater and remains so through 2017. It appears that other enviornmental factors may play more of a factor then just forest condition alone.
Another important trend we want to examine is if there is a significant effect of forest condition (old-growth or clearcut) on where in the channel salamanders are found (channel classification). To do so we used location data on where salamanders were observed in Mack Creek in 2017.
Figure 5. Number of Pacific giant salamanders recorded in Mack Creek in 2017 by creek section (old-growth/clearcut) and channel classification.| Channel Class | Clear Cut | Old Growth |
|---|---|---|
| Cascade | 247 (55%) | 201 (45%) |
| Pool | 31 (41%) | 45 (59%) |
| Side Channel | 90 (55%) | 74 (45%) |
Our analysis found that there is no significant association between forest condition (old-growth/clearcut) and where salamanders are found in the creek (pool/side-channel/cascade) in 2017 by a chi-squared test: (\(\chi\)2(2) = 5.54, p = 0.063).
In addition to analyzing the effect forest condition has on abundance, we want to see if it also has any effect on Pacific giant salamander size on average. To do so, we looked at recorded weights for these salamanders in 2017.
Figure 6. Recorded weights of Pacific giant salamanders sampled in Mack Creek in 2017 by section of the creek found (old-growth & clearcut). The black dot represents the sample mean with error bars representing the range of uncertainty. Data: Andrews Forest LTER
We can see that the majority of salamanders sampled in 2017 for both sections of the creek had a weight under 15 grams. Also notable are the several outliers in both the clearcut and old-growth sections of the creek that were signifincatly larger than most of the individuals in the samples.
Figure 7. Pacific giant salamander sample statistics in 2017 by section of Mack Creek (old-growth/clearcut)| Creek Section | Mean Weight (g) | Standard Deviation (g) | Sample Size | Standard error | Variance |
|---|---|---|---|---|---|
| Clear Cut | 7.78 | 9.90 | 368 | 0.52 | 98.1 |
| Old Growth | 6.58 | 8.96 | 328 | 0.49 | 80.2 |
In 2017, Pacific giant salamanders in the clearcut section of Mack Creek were on average 1.2 grams (18%) larger then those sampled in the old-growth section. While the sample data is skewed to the left with several outliers affecting the mean, the two sample sizes are quite large. In addition, the variances were found to be equal by a Levene’s test (p = 0.29). For these reasons, we justified using a paramatric test to compare means between the two samples as being appropriate.
Our analysis found that there is not enough evidence to conclude that mean Pacific giant salamander size in the clearcut section of Mack Creek in 2017 (7.78 \(\pm\) 9.9, n = 368) differed significantly from mean Pacific giant salamander size in the old-growth section of Mack Creek in 2017 (6.58 \(\pm\) 8.96, n = 328) by a two-sided, two sample t-test (t(692.79) = -1.67, p = 0.096). In addition, the effect size between mean sizes is negligible (Cohen’s d = 0.13).
The uncertainty of precipitation patterns in the future could result in the size and flow of creeks in the Pacific Northwest drastically changing. If water flow is reduced this may lead to an increase in pool microhabitats and a loss in microhabitats such as cascades within a given creek. To test the significance this might have to Pacific giant salamanders it is important to see whether the species shows signs of increased vitality in different creek classifications. To do so, we looked at mean salamander weights in three different creek classifications (pool, cascade, side channel) within Mack Creek in 2017.
Figure 8. Recorded weights of Pacific giant salamanders sampled in Mack Creek in 2017 by channel classification. The black dot represents the sample mean with error bars representing the range of uncertainty.
We can see that the majority of salamanders sampled in 2017 for all three channel classifications also had a weight under 15 grams. Again, there are notable outliers that were significantly larger than most of the individuals in the sample.
Figure 9. Pacific giant salamander sample statistics in 2017 by different channel classifications within Mack Creek| Channel Class | Mean Weight (g) | Standard Deviation (g) | Sample Size | Standard error | Variance |
|---|---|---|---|---|---|
| Cascade | 7.52 | 9.03 | 448 | 0.43 | 81.50 |
| Pool | 9.30 | 13.62 | 76 | 1.56 | 185.58 |
| Side Channel | 5.68 | 8.27 | 164 | 0.65 | 68.31 |
In 2017, Pacific giant salamanders sampled within pools in Mack Creek were the largest. On average, the pool salamanders were 1.78 grams (20%) larger then those found in cascades and 3.68 grams (65%) larger then those found in side-channels. Also notable is the number of salamanders found in pools was about 1/2 of those found in side-channels and nearly 1/6 of those found in cascades.
The sample data for salamander weight by channel class is similarly skewed to the left with several outliers affecting the mean. However, the sample size is adequately large and the variances were found to be equal by a Levene’s test (p = 0.09). For these reasons, we used a paramatric test to compare means between the three variables.
Our analysis found that mean salamander weight (g) differed significantly between the three creek channel classifications (pool, cascade, side-channel) by a one-way ANOVA test (F(2, 684) = 4.22, p = 0.015). Further pairwise testing using pot-hoc Tukey’s HSD revealed that mean salamander weight was only significantly different between side-channel and pool classifications (p = 0.017). Though a significant difference in mean weight was found between salamanders sampled in the pools vs. side-channel sections of Mack Creek in 2017, the effect size was small (Cohen’s d = 0.35).
Pacific giant salamander populations, despite logging, have remained constant between 1993 and 2017. Looking at the trend lines in Figure 4, it is apparent that regardless of forest condition, clear cut or old growth, the salamander population has thrived.
After visually and analytically exploring salamander weights between the two forest conditions, it is apparent, given Figures 7 and 8, that there is not enough evidence to conclude that forest condition affects salamanders’ weights; this further supports the notion that logging has had a limited effect on salamander health and population.
Although salamander population size and weight do not differ between old-growth and clearcut forests, it is apparent, given Figure 9, that channel class did have a significant effect on salamander weight. The “pool” class proved to have salamanders with the largest mean weight when compared to cascades and side channels. Even though forest condition may not directly affect salamander weights, perhaps there is an unforeseen factor that is causing salamanders to grow larger in pools over side channels or cascades. Further research would need to be done to determine the reasons behind this observation.
Gregory S. V. 2016. Aquatic Vertebrate Population Study in Mack Creek, Andrews Experimental Forest, 1987 to present. Environmental Data Initiative. https://doi.org/10.6073/pasta/5de64af9c11579266ef20da2ff32f702. Dataset accessed 11/27/2019.
“About the Andrews Forest.” HJ Andrews Experimental Forest, https://andrewsforest.oregonstate.edu/about.
“Pacific Giant Salamander.” Wildlife Profiles, Oregon Wild, https://oregonwild.org/wildlife/pacific-giant-salamander.